Learning Visual Question Answering by Bootstrapping Hard Attention
This addresses the need for more efficient attention mechanisms in computer vision, particularly for tasks like visual question answering, though it is incremental as it builds on existing attention concepts.
The paper tackled the problem of hard attention in visual question answering by introducing a new approach that selectively ignores some features, achieving competitive performance on recent datasets and equaling or surpassing soft attention architectures while being more efficient.
Attention mechanisms in biological perception are thought to select subsets of perceptual information for more sophisticated processing which would be prohibitive to perform on all sensory inputs. In computer vision, however, there has been relatively little exploration of hard attention, where some information is selectively ignored, in spite of the success of soft attention, where information is re-weighted and aggregated, but never filtered out. Here, we introduce a new approach for hard attention and find it achieves very competitive performance on a recently-released visual question answering datasets, equalling and in some cases surpassing similar soft attention architectures while entirely ignoring some features. Even though the hard attention mechanism is thought to be non-differentiable, we found that the feature magnitudes correlate with semantic relevance, and provide a useful signal for our mechanism's attentional selection criterion. Because hard attention selects important features of the input information, it can also be more efficient than analogous soft attention mechanisms. This is especially important for recent approaches that use non-local pairwise operations, whereby computational and memory costs are quadratic in the size of the set of features.